Upload
others
View
7
Download
0
Embed Size (px)
Citation preview
Channel change and ¯ooding, Skokomish River, Washington
S.C. Stover*, D.R. Montgomery1
Department of Geological Sciences, University of Washington, Seattle, WA 98195, USA
Received 3 February 2000; revised 13 December 2000; accepted 15 December 2000
Abstract
Analysis of 45 years of cross-section data documents changes in channel width and bed elevation along the Skokomish River,
Washington. The bed of the South Fork Skokomish incised over 1 m between 1940 and 1964, although both prior and
subsequent to this period the mean bed elevation oscillated as much as 0.8 m with almost no cumulative change. In contrast,
the mainstem Skokomish channel bed aggraded nearly 0.5 m between 1939 and 1944, oscillated at amplitudes up to 1 m with
little net change from 1945 to 1964, and aggraded over 1.3 m between 1965 and 1997. In the late 1920s, prior to the onset of
gaging station records, damming of the North Fork signi®cantly reduced ¯ow in the mainstem Skokomish. From the 1930s to
the 1990s, peak discharge data for both the South Fork and the mainstem indicate no net increase in peak ¯ows. Despite the
reduced discharge from dam construction in the 1920s and no increase in peak ¯ows during the following years, the frequency
of overbank ¯ooding in recent decades has increased on the ¯oodplain of the mainstem. Systematic written descriptions and
aerial photographs of the catchment from 1929 to 1992 document land use, including timber harvesting, road construction, and
in-channel debris removal. The timing of changes in channel width and elevation imply that debris removal may have triggered
periods of degradation and that near-channel and headwater land use potentially elevated the sediment supply to both reaches.
Although direct land use causality is dif®cult to constrain, progressive reduction of channel conveyance in the mainstem as
observed in USGS gage height trends does indicate that increased ¯ooding on the mainstem Skokomish River resulted from
aggradation, without an increase in peak discharges. q 2001 Elsevier Science B.V. All rights reserved.
Keywords: Land use; Sediment transport; Channel geometry; Floods; Rivers; Olympic Peninsula
1. Introduction
The potential for land management to in¯uence
downstream ¯ooding has been broadly acknowledged
since early investigators discussed the effects of vege-
tation removal on erosion and runoff generation
(Surell, 1841; Marsh, 1864). Although it is widely
recognized that urbanization-related changes in
basin hydrology can substantially alter ¯ood
frequency (e.g. James, 1965; Hollis, 1975; Booth,
1991; Moscrip and Montgomery, 1997), the effect of
rural land management on downstream ¯ooding
remains more controversial. Increased ¯ooding
could arise from either hydrologic changes that
increase peak ¯ows or channel changes that reduce
conveyance, and thereby trigger greater ¯ooding
without directly changing discharge quantity or
timing. Most studies indicate that water yields
increase after forest clearing (Bosch and Hewlett,
1982), but it is not clear as to whether industrial
forestry signi®cantly affects downstream ¯ooding.
The hydrologic response to timber harvest increases
Journal of Hydrology 243 (2001) 272±286www.elsevier.com/locate/jhydrol
0022-1694/01/$ - see front matter q 2001 Elsevier Science B.V. All rights reserved.
PII: S0022-1694(00)00421-2
* Corresponding author. Department of Geological Sciences,
University of Colorado, Boulder, CO 80309, USA. Tel.: 11-303-
735-5032; fax: 11-303-492-2606.
E-mail addresses: [email protected] (S.C. Stover),
[email protected] (D.R. Montgomery).1 Tel.: 11-206-685-2560; fax: 11-206-543-3836.
small, early season high ¯ows, but researchers have
not found consistently detectable effects on larger,
¯ood-generating ¯ows (Rothacher, 1970; Rothacher,
1973; Harris, 1973; Ziemer, 1981; Lyons and Beschta,
1983; Jones and Grant, 1996; Thomas and Megahan,
1998). In contrast to the past emphasis on changes in
peak ¯ows, little attention has been focused on the
role of channel change in downstream ¯ooding in
forested mountain drainage basins. Here we use U.S.
Geological Survey (USGS) gaging station records,
supplemented with historical aerial photographs, to
examine long-term channel change and its in¯uence
on ¯oodplain inundation during a period of intensive
upland management in the drainage basin of the
Skokomish River, Washington. Analysis of frequent
discharge sampling and cross-section surveys demon-
strates that reduced channel conveyance has led to
more frequent ¯oodplain inundation despite reduced
peak ¯ows.
2. Study area
The Skokomish River drains 622 km2 of forested
terrain on the leeward side of the Olympic Peninsula,
Washington (Fig. 1). The catchment discharges
directly into Hood Canal, and much of the northern
perimeter abuts Olympic National Park. Relief in the
catchment is 424 m, and annual precipitation averages
3400 mm, most of which falls from November to
February (Canning et al., 1988). The Skokomish
catchment is underlain by thick deposits of glacial
till and inter-glacial gravel, which is, in turn, under-
lain by basalt (Carson, 1970; Tabor, 1975). The lower
portions of the basin consist of till deposited by the
Vashon lobe of the Fraser Glaciation approximately
15,000 B.P. In the valley of the mainstem Skokomish,
interglacial gravel inter®ngers with Holocene ¯uvial
deposits of the Kitsap Formation. Eocene basalt
outcrops in upper portions of the basin, in the gorge
S.C. Stover, D.R. Montgomery / Journal of Hydrology 243 (2001) 272±286 273
USGS Gage12060500
USGS Gage12061500
Cushman Dam
Lower CushmanDam
North ForkSkokomish
South ForkSkokomish
MainstemSkokomish
LakeCushman
LakeKokanee
5 km
Fig. 1. General study area map outlining the Skokomish catchment and the locations of the South Fork and mainstem USGS gaging stations.
The channel gradient decreases from 0.007 at the South Fork gaging station to less than 0.005 at the mainstem gaging station.
immediately above the lower South Fork gaging
station, and in isolated locations along the mainstem.
The Skokomish catchment has an extensive history
of land management. Beginning in the late 1800s, the
Skokomish lowlands were transformed into dairy
farms and agricultural land. Near-channel timber
harvesting started in the late 1800s in the lowlands
and progressed into the uplands by the 1940s. By
1992, 80% of the drainage basin of the South Fork
had been harvested and thousands of miles of logging
roads and railroad lines had been constructed in the
catchment (KCM, 1993; Canning et al., 1988). Other
practices, including gravel mining in the 1930s,
channel straightening in 1953, and dike and berm
construction circa 1971 also in¯uenced the Skokomish
River. In addition, construction of the Cushman and
Lower Cushman dams in the late 1920s on the North
Fork diverted approximately 40% of the mainstem
¯ow directly to Hood Canal (Williams et al., 1985;
Jay and Simestad, 1994). Despite overall ¯ow reduc-
tion due to the dams, recurrent inundation of the ¯ood-
plain of the mainstem Skokomish River has sparked
controversy over the role of upland land management
practices in ¯ooding along the mainstem Skokomish.
3. Methods
We analyzed the entire available suite of USGS
discharge data for the South Fork gaging station
(#12060500) and the mainstem station (#12061500)
(Fig. 1) in order to evaluate temporal changes in chan-
nel morphology. Original discharge measurement
sheets (USGS Form 9-275-F) and supplementary
discharge notes (USGS Form 9-275d) were used to
develop cross-sections of the channel and to analyze
other evidence of channel change recorded by USGS
personnel. The South Fork and mainstem historically
have been sites of geologic and aquatic habitat inter-
est, and discharge measurements were taken on nearly
a monthly basis throughout the period of record. The
unusual frequency of measurements and long periods
of record result in a substantial data set, with a total of
499 discharge surveys at the South Fork and 544
surveys at the mainstem. All discharge measurements
post-date construction of the Cushman and Lower
Cushman dams, and hence the mainstem ¯ow record
re¯ects the reduction in ¯ow due to dam diversion.
The data record for the South Fork extends from
1932 to 1984 and the location of the gaging station
was not altered during this period. Hence, all measure-
ments are referenced to a consistent datum. At the
mainstem gaging station, the available period of
record covers 1932 to 1997. During this period, the
gaging station was relocated three times. From 1956
to 1965, the gage location oscillated between the
original gaging station and a location directly across
the channel because ®ne sediment periodically
restricted the intake valves. Measurements during
this period were made at whichever gaging station
exhibited clear intake valves. In July 1965, the gage
was permanently relocated downstream at the High-
way 101 bridge. Each new site was located within a
kilometer from the original gaging point, and well-
documented survey data and differences in elevation
were used to correlate the data sets from each location
of the mainstem gage.
3.1. Channel geometry determination
Each discharge estimate by USGS personnel was
based on depth measurements at intervals across the
channel and a gage height read from the station gage.
From these data, we employed the following proce-
dure to calculate mean bed heights, mean thalweg
heights, and channel width. Both cable and wading
measurements were collected at each site, with
wading locations documented in terms of distance
from the gaging station and a local control (i.e. stable
large barform). As the control location may have
changed with ¯ow conditions, the distance above or
below the gaging station, as noted on the measure-
ment sheets, was used to select those measurements
that could be considered consistent with the cable
measurements. If the measurement location exceeded
200 ft (60.8 m) above or below the gage as documen-
ted on the measurement sheet, then the measurement
was not used. For each discharge measurement, a
channel depth cross-section was constructed by
subtracting the gage height from total depth at each
cross-sectional point. Data points from either bank
were removed from each cross-section data set to
restrict the analysis to variations in channel bed eleva-
tion. The data sets for each measurement were then
used to calculate mean bed elevation and channel
width. These parameters and gage height/discharge
S.C. Stover, D.R. Montgomery / Journal of Hydrology 243 (2001) 272±286274
relationships were evaluated for changes over time.
Distinct phases of channel behavior were identi®ed
through changes in both long-term trends in mean
bed elevation and the amplitude of ®ner-scale
oscillations.
3.2. Statistical analysis of width changes
Paired t-tests were used to evaluate the signi®cance
of width changes for the South Fork and mainstem,
following Beschta (1983) and testing for signi®cance
at the a � 0:05 con®dence level. To isolate times of
distinct channel change, periods of statistically signif-
icant width change were subdivided and paired t-tests
were reconducted. Following Hoey (1994), we also
implemented an analytical approach to identify peri-
ods of aggradation and degradation for the two chan-
nels. This approach uses distance between channels
and width changes over time to calculate aggradation
and degradation magnitudes. Volumetric changes
(DV), and hence aggradation and degradation trends,
between the South Fork and mainstem reaches were
determined through using the following equation and
assuming that D�zand wi vary linearly between the
reaches:
DV � wiD�ziL�2 1 a 1 b 1 2ab�=6 �1�where wi is the bank-to-bank width (m) at reach i; D�zi
is DA/w (m); A is cross-sectional area (m2); L is the
distance between reaches i and j (m); a is wi/wj; and b
is D�zi=D�zj:
Volumetric changes were converted to elevation
changes for presentation clarity following:
D �hi � DV
wiLB�2�
where D �hi is average elevation change (m) at reach i,
and B is �2 1 a 1 b 1 2ab�=6:Changes were computed over ten-year intervals to
assess long-term trends in aggradation and degrada-
tion. Cross-sectional area and width data were
extracted from each discharge sheet and averaged
for each period of interest. Distance data were updated
for the time periods in which the mainstem gaging
station was relocated, with distance between the
reaches measuring 8280 m from 1932 to 1965 and
9240 m from 1965 to the end of the record. Absolute
aggradation and degradation values also were
determined through time trends in the mean bed
heights for each reach over parallel periods of interest.
3.3. Characterization of land management
The continuous record of discharge measurements
and supplementary discharge notes record changes in
land use of the Skokomish catchment. Impressively
systematic written descriptions in the remarks section
of the discharge measurements (i.e. 88% of South
Fork forms and 73% of mainstem forms contained
written documentation), and additional notation on
supplementary discharge notes document near-
channel timber harvest, bedform changes, and in-
channel debris removal near the gage locations. The
nearly monthly documentation associated with the
discharge measurements constrained the timing of
many of the land use activities in the vicinity of the
gaging stations. To supplement the written description
of changes in land use, we evaluated harvested areas,
land use changes, and road construction apparent on
available sets of aerial photographs of the catchment
(1929, 1944, 1952, 1965, 1981, and 1992). The timing
of land use activities determined through both written
descriptions and aerial photo evaluation aided inter-
pretation of trends in the mean bed elevation.
4. South Fork channel change
Changes in bed elevation de®ne three distinct
phases of channel response for the South Fork (Fig.
2). The ®rst phase (1932±1944) is de®ned by initial
slight aggradation and moderate-amplitude ¯uctua-
tions in mean bed elevation from 1932 to 1940,
followed by minor incision from 1940 to 1944. The
second phase of response (1944±1964) began with
low-magnitude aggradation (1944±1950) followed
by subsequent degradation from 1951 to 1964. The
®nal phase of bed response (1964±1984) suggests
stabilization of mean bed height. Land management
activity during these three phases provides insight into
the potential causes for each stage of channel response.
4.1. 1932±1944
From 1932 to 1940 the channel was characterized
by a fairly stable mean bed elevation, but from 1940 to
1944 the channel incised (Fig. 2). From 1932 to 1935,
S.C. Stover, D.R. Montgomery / Journal of Hydrology 243 (2001) 272±286 275
the mean bed height aggraded slightly and ¯uctuated
with approximately 0.6 m oscillations. Throughout
this time period USGS personnel documented an
in¯ux of ªloose gravelº into the channel and noted
that ªdrift [wood debris] comes down oftenº (USGS
Discharge Sheets, various dates, 1932 to 1934). High
¯ows between 1933 and 1935 may have initiated inci-
sion, but the bed elevation quickly recovered to a pre-
¯ood elevation (Fig. 3). Land use activities from 1932
to 1935, as noted on USGS discharge sheets, consisted
primarily of near-bank timber harvesting along the
South Fork, near and upstream of the gaging station.
Between 1935 and 1939, the channel exhibited seaso-
nal ®ll and scour, and the amplitude of the bed
oscillations decreased to about 0.3 m. By early
1939, USGS personnel had begun to note ªpiles of
logsº on the section control (USGS Discharge Sheet,
January 12, 1939), and in late 1939, there was docu-
mentation of a ªriver full of logs and debrisº (USGS
Discharge Sheet, November 24, 1939), suggesting an
increased amount of woody debris in the channel. A
December 1939 high ¯ow event and in-channel debris
removal in early 1940 correspond to a time of net bed
lowering (Figs. 2 and 3).
The near-channel disturbances continued into 1940,
and channel response consisted of minor aggradation
followed by degradation beginning in mid-1941.
USGS personnel noted ®ne-grained sediment input
early in 1940 and ªloose gravelº later that same
year; discharge sheets document and aerial photos
support the presence of landslides ªin vicinity of
gageº (USGS Discharge Sheet, March 3, 1941) and
in-channel logjams from 1940 to mid-1941. Proximal
disturbances consisted of near-channel timber
harvesting ªimmediately above gageº (USGS
Discharge Sheets, September 5, 1940 and October
S.C. Stover, D.R. Montgomery / Journal of Hydrology 243 (2001) 272±286276
30
30.5
31
31.5
32
32.5
33
33.5
34
1930 1940 1950 1960 1970 1980
Year
Ele
vati
on
(mA
NG
VD
)
South Fork Mean Bed Height
PHASE 1
PHASE 2
PHASE 3
1932 -1935Near -
channelHarvesting
1939 - 1940Removal ofIn-channel
Debris1940 -1941
Near-channelHarvesting
and In-channel
Logging Road
1949 - 1961High Flow Log Mobilization
and Physical Log HaulingAcross Channel
1951 - 1961Extensive Upstream and Left-
bank Harvesting1964 -1984
Headwater Harvesting and Road Construction30
30.5
31
31.5
32
32.5
33
33.5
34
1930 1940 1950 1960 1970 1980
Year
Ele
vati
on
(mA
NG
VD
)
South Fork Mean Bed Height
PHASE 1
PHASE 2
PHASE 3
1932 -1935Near -
channelHarvesting
1939 - 1940Removal ofIn-channel
Debris1940 -1941
Near-channelHarvesting
and In-channel
Logging Road
1949 - 1961High Flow Log Mobilization
and Physical Log HaulingAcross Channel
1951 - 1961Extensive Upstream and Left-
bank Harvesting1964 -1984
Headwater Harvesting and Road Construction
Fig. 2. Temporal changes in mean bed elevation, phases of channel response, and general land management activities for the South Fork
Skokomish River. ANGVD is above national geodetic vertical datum.
16, 1940), and the late 1941 development of an in-
channel logging road. The trend in mean bed height
indicates channel degradation from late 1941 to 1944,
with the mean bed elevation decreasing by about
0.3 m. This degradation corresponds to the frequent
documentation of ªrunning driftº on USGS discharge
sheets. Throughout the initial aggradation and latter
degradation during this response phase, the change in
channel width was statistically insigni®cant (Fig. 4a).
4.2. 1944±1964
Channel response from 1944 to 1964 was charac-
terized by gradual channel incision (Fig. 2). A trend of
lowered bed elevation began in a high ¯ow event
during 1949 (Fig. 3), which USGS personnel docu-
mented as introducing ®ne sediment and wood debris
to the reach. This trend continued until 1964, with a
marked increase in bed height oscillations from 1951
to 1961. Such change in the system corresponds to a
period of upstream and extensive local left bank
timber harvesting, in the immediate vicinity of the
gage. Both written USGS documentation (USGS
Discharge Sheets, various dates 1951 to 1961) and
aerial photographs indicated increased timber harvest-
ing from 1944 to 1965. USGS personnel further noted
a high-water event that mobilized logs in the channel
in 1954 and loggers dragging logs across the channel
in 1956. Throughout the period of degradation from
1950 to 1964, the mean bed height decreased over
1 m, and the channel width increased more than
20% (Figs. 2 and 4a).
4.3. 1964±1984
From 1964 to 1984, the channel exhibited no
distinct aggradation or degradation (Fig. 2), despite
three distinct peak ¯ow events in excess of 400 cms
(Fig. 3). Bed height oscillations of 0.3±0.8 m
continued, although no signi®cant change in width
S.C. Stover, D.R. Montgomery / Journal of Hydrology 243 (2001) 272±286 277
0
100
200
300
400
500
600
700
1932
1933
1936
1937
1939
1941
1943
1946
1947
1949
1952
1954
1955
1957
1959
1962
1963
1966
1968
1970
1972
1974
1975
1977
1979
1981
1983
Year
Pea
kD
isch
arg
e(c
ms)
30
30.5
31
31.5
32
32.5
33
1931 1941 1951 1961 1972 1982
Year
Ele
vati
on
(m)
Peak Discharge (cms)
South Fork Mean Bed Height (m)Phase One
Phase Two
Phase Three
0
100
200
300
400
500
600
700
1932
1933
1936
1937
1939
1941
1943
1946
1947
1949
1952
1954
1955
1957
1959
1962
1963
1966
1968
1970
1972
1974
1975
1977
1979
1981
1983
Year
Pea
kD
isch
arg
e(c
ms)
30
30.5
31
31.5
32
32.5
33
1931 1941 1951 1961 1972 1982
Year
Ele
vati
on
(m)
Peak Discharge (cms)
South Fork Mean Bed Height (m)Phase One
Phase Two
Phase Three
Fig. 3. Temporal changes in mean bed elevation superimposed on peak discharge values for the South Fork Skokomish River.
was noted for this period (Fig. 4a). Throughout this
time, no near-bank logging was documented on the
USGS discharge sheets, and the main land use
activity, as observed in 1965 and 1981 aerial photos
and recorded in USGS discharge sheets, consisted of
upland road construction and timber harvesting.
5. Mainstem channel change
Changes in channel geometry for the mainstem
Skokomish are well described by the same three
time periods that de®ne phases of channel response
on the South Fork (Figs. 5 and 6). The ®rst phase
(1932±1944) exhibited both an initial incision from
1932 to 1938, and subsequent aggradation from 1939
to 1944. The second phase (1944±1964) was charac-
terized by bed elevation ¯uctuations of 0.3±1 m in
amplitude. The ®nal phase of response (1964±1997)
involves accelerated aggradation, and a dampening of
the oscillations in mean bed height. Evidence of land
management activity during these three phases of
response provides insight into the potential causes of
changes in channel width and bed elevation.
S.C. Stover, D.R. Montgomery / Journal of Hydrology 243 (2001) 272±286278
-10
-5
0
5
10
15
20
25
30
1932-1944 1944-1964 1964-1984
Time Period (years)
Per
cen
tWid
thC
han
ge
α = 0.01
NS NS
-4
-2
0
2
4
6
8
10
12
14
16
1932-1944 1944-1964 1964-1997
Time Period (years)
Per
cen
tWid
thC
han
ge
NS
NS
α = 0.01
(a)
(b)
-10
-5
0
5
10
15
20
25
30
1932-1944 1944-1964 1964-1984
Time Period (years)
Per
cen
tWid
thC
han
ge
α = 0.01
NS NS
-4
-2
0
2
4
6
8
10
12
14
16
1932-1944 1944-1964 1964-1997
Time Period (years)
Per
cen
tWid
thC
han
ge
NS
NS
α = 0.01
-10
-5
0
5
10
15
20
25
30
1932-1944 1944-1964 1964-1984
Time Period (years)
Per
cen
tWid
thC
han
ge
α = 0.01
NS NS
-4
-2
0
2
4
6
8
10
12
14
16
1932-1944 1944-1964 1964-1997
Time Period (years)
Per
cen
tWid
thC
han
ge
NS
NS
α = 0.01
(a)
(b)
Fig. 4. Signi®cance of percent width change for the: (a) South Fork; and (b) mainstem Skokomish. NS denotes no signi®cant change; a denotes
level of signi®cance. (a) Width change was signi®cant to a � 0:01 for 1944±1964, and insigni®cant for 1932±1944 and 1964±1984. (b) Width
change was signi®cant to a � 0:01 for 1964±1997, and insigni®cant for 1932±1944 and 1944±1964.
5.1. 1932±1944
Initial response on the mainstem involved incision
of nearly 0.5 m from 1932 to 1944 (Fig. 5). This rapid
incision spans a period of intense channel-proximal
disturbances: 1932±1934 gravel mining at the gage,
and along the left bank abutting the gage, for construc-
tion of Highway 101; and the 1934 construction of the
highway (which introduced gravel and forced the
upstream relocation of USGS channel control
(USGS Discharge Sheet, June 5, 1934)). From 1939
to 1944 the channel aggraded to nearly its original
1932 elevation. Predominant land use during the
1939 to 1944 period consisted of localized near-
bank timber harvesting. During this period, the
change in channel width was insigni®cant (Fig. 4b).
5.2. 1944±1964
The period from 1944 to 1964 was characterized by
an enhanced amplitude of bed surface ¯uctuations,
although the mean bed elevation throughout Phase
Two did not change appreciably (Fig. 5). This phase
of channel response coincides with the onset of wide-
spread upstream timber harvesting and road develop-
ment within the catchment. In 1947, the channel also
began to exhibit bed elevation oscillations of up to
1 m and USGS notes state that the channel bed
began to ®ne. The extreme bed elevation ¯uctuations
continued until 1964. In 1953, channel straighten-
ing eliminated the bend upstream of the gage.
Width change during this period was insigni®cant
(Fig. 4b).
S.C. Stover, D.R. Montgomery / Journal of Hydrology 243 (2001) 272±286 279
3
3.5
4
4.5
5
5.5
6
6.5
7
1930 1940 1950 1960 1970 1980 1990
Year
Ele
vati
on
(mA
NG
VD
)
Mainstem Mean Bed Height
PHASE 1
PHASE 2
PHASE 3
1926Cushman/
1930 LowerCushman
DamInstallation
1953Channel Straightening
1944-1964Widespread Upstream Harvesting and
Road Development
1932-1934GravelMining
1935Isolated
Harvesting
1939-1942Near-bankHarvesting
1964-1997Headwater Harvesting and Road Construction
1971-1975Riprap placement
3
3.5
4
4.5
5
5.5
6
6.5
7
1930 1940 1950 1960 1970 1980 1990
Year
Ele
vati
on
(mA
NG
VD
)
Mainstem Mean Bed Height
PHASE 1
PHASE 2
PHASE 3
1926Cushman/
1930 LowerCushman
DamInstallation
1953Channel Straightening
1944-1964Widespread Upstream Harvesting and
Road Development
1932-1934GravelMining
1935Isolated
Harvesting
1939-1942Near-bankHarvesting
1964-1997Headwater Harvesting and Road Construction
1971-1975Riprap placement
Fig. 5. Temporal changes in mean bed elevation, phases of channel response, and general land management activities for the mainstem
Skokomish River. ANGVD is above national geodetic vertical datum.
5.3. 1964±1997
From 1964 to 1997 the mainstem aggraded and
USGS discharge forms document signi®cant channel
®lling and the formation of a soft channel substrate
from ®ning with ªmudº. Total elevation gain was over
1.3 m and the channel width increased by over 12%
(Figs. 4b and 5). This aggradation eventually forced
relocation of the gaging station, as the intake valves
were beginning to become isolated from the main
channel. Between 1971 and 1975, riprap and natural
berm material were emplaced along the right bank of
the channel, upstream of the gaging location (USGS
discharge sheets, various). Perhaps coincidentally, the
distinct bed-height oscillations ceased after 1976.
Land use practices during this phase consisted of
upland timber harvesting and road construction,
primarily upstream of the South Fork gaging
station.
6. Peak ¯ows, stage±discharge relation and¯oodplain inundation
Evaluation of peak discharge recurrence intervals
and stage height for each phase of channel response in
both the South Fork and mainstem indicates that the
recent ¯ooding along the mainstem is a result of an
altered stage±discharge relation. Analysis of peak
discharge recurrence intervals for each phase of chan-
nel response reveals that the peak discharge is either
decreasing over time, or remaining relatively
unchanged for both reaches (Fig. 7). Although neither
the South Fork nor the mainstem exhibit an increase in
peak discharge recurrence intervals throughout the
period of record, both reaches display systematic
changes in the stage, or water surface elevation as
re¯ected in gage height measurements, at any given
discharge. For the South Fork, the ¯ow stage±
discharge relation indicates a lowering of the water
S.C. Stover, D.R. Montgomery / Journal of Hydrology 243 (2001) 272±286280
0
200
400
600
1000
1200
1943
1946
1947
1949
1952
1954
1955
1957
1959
1961
1963
1966
1968
1970
1972
1974
1975
1977
1979
1982
1983
1986
1987
1989
1992
1993
Year
Pea
kD
isch
arg
e(c
ms)
0
1
2
3
4
5
7
1942 1952 1962 1972 1982 1992
Year
Ele
vati
on
(m)
Peak Discharge (cms)
Mainstem Mean Bed Height (m)Phase Two Phase Three
0
200
400
600
800
1000
1200
1943
1946
1947
1949
1952
1954
1955
1957
1959
1961
1963
1966
1968
1970
1972
1974
1975
1977
1979
1982
1983
1986
1987
1989
1992
1993
Year
Pea
kD
isch
arg
e(c
ms)
0
1
2
3
4
5
6
7
1942 1952 1962 1972 1982 1992
Year
Ele
vati
on
(m)
Peak Discharge (cms)
Mainstem Mean Bed Height (m)Phase Two Phase Three
Fig. 6. Temporal changes in mean bed elevation superimposed on peak discharge values for the mainstem Skokomish River. Due to a lack of
continuous peak discharge data from 1932 to 1943, the time period presented here covers 1943±1993.
surface elevation for a given discharge until the
stage±discharge relationship stabilized in 1964 (Fig.
8). This trend parallels the entrenchment of the South
Fork channel bed and the beginning of apparent bed
elevation stability in 1964. In contrast, the mainstem
¯ow stage versus discharge relationship exhibits an
increased stage for any given discharge, coinciding
with an aggrading channel bed (Fig. 9). Such reduc-
tion in the requisite discharge for a given stage,
together with a relatively unchanged peak discharge
regime, indicates that recent mainstem ¯ooding is due
to aggradation within the channel. Without simulta-
neous aggradation of the ¯oodplain, the higher stage
associated with smaller discharges allows for a higher
frequency of ¯oodplain inundation on the mainstem.
7. Discussion
The South Fork and mainstem reaches of the
Skokomish River exhibited different styles of
response during the period of record. The South
Fork channel initially incised, and then stabilized its
mean bed elevation after 1964, although signi®cant
bed-elevation ¯uctuations continued. In contrast, the
mainstem experienced a short period of degradation in
the early 1930s, after which the channel recovered its
initial elevation and began to exhibit signi®cant bed
oscillations. After 1964, these oscillations gave way
to gradual aggradation, which totaled over 1.3 m by
1997.
The parallel timing of South Fork debris removal
and in-channel perturbations with degradation of the
bed suggests that the in-channel events caused the
observed South Fork channel response. In contrast,
the presence of the Cushman and Lower Cushman
dams, which were built prior to installation of the
gaging stations, raises the possibility that aggradation
along the mainstem resulted from dam construction.
Such causality hinges on the idea that reduced ¯ow to
the mainstem lowered the transport capacity of the
mainstem reach. However, it is dif®cult to ascribe
the aggradation of the mainstem solely to reduced
discharge from the dammed North Fork, because
damming should have caused mainstem aggradation
relatively quickly. This is especially true given the
comparatively low overall discharge regime from
1932 to 1944, the ®rst recorded period of post-dam
discharge. It is also dif®cult to attribute observed
channel changes to altered peak ¯ow frequency, as
the discharge data show no discernible increase in
peak ¯ows for either reach. We infer that alterations
in sediment load from proximal and distal harvesting-
related activities and road construction, followed by
S.C. Stover, D.R. Montgomery / Journal of Hydrology 243 (2001) 272±286 281
(a) (b)
100
1000
1 10 100
Peak Q (cms) (1932-1944)
Peak Q (cms) (1944-1964)
Peak Q (cms) (1964-1984)
100
1000
1 10 100
Peak Q (cms) (1944-1964)
Peak Q (cms) (1964-1997)
Peak
Q( c
ms)
Peak
Q( c
ms)
Recurrence Intervals (Years) Recurrence Intervals (Years)
(a) (b)
100
1000
1 10 100
Peak Q (cms) (1932-1944)
Peak Q (cms) (1944-1964)
Peak Q (cms) (1964-1984)
100
1000
1 10 100
Peak Q (cms) (1944-1964)
Peak Q (cms) (1964-1997)
Peak
Q( c
ms)
Peak
Q( c
ms)
Recurrence Intervals (Years) Recurrence Intervals (Years)
Fig. 7. Recurrence Interval vs. Peak Discharge for the: (a) South Fork; and (b) mainstem Skokomish. (a) The trends in recurrence interval over
time indicate a generally decreasing discharge regime for low and high recurrence intervals. (b) The trends in recurrence interval over time
indicate a relatively unchanged discharge regime for low recurrence intervals and a decrease in the discharge regime for high recurrence
intervals.
debris removal from land management, triggered the
observed South Fork channel response. Likewise, we
infer that an increased sediment load from proximal
harvesting, upstream harvest-related activities, and
South Fork incision, combined with a reduced trans-
port capacity from dam construction, were respon-
sible for the observed mainstem channel response.
7.1. South Fork
The South Fork generally incised from 1932 to the
mid-1950s, and subsequently exhibited distinct oscil-
lations in bed elevation until the 1960s. The presence
of wood debris in stream channels may modify
hydraulic properties and bedforms, and removal of
debris can result in dramatic and rapid reduction in
bed elevation and hydraulic roughness (Macdonald
and Keller, 1987; Shields and Smith, 1992; Smith et
al., 1993a,b). We suspect that harvesting of riparian
forests in conjunction with the removal of wood
debris from the South Fork, together with a late
1939 high ¯ow event, initiated channel incision
from late 1939 to 1940. In the second, more extensive
period of degradation (late-1940±1944), the drop in
channel elevation coincided with debris removal at
the gage location to clear the channel for development
of a temporary in-channel logging road. During this
same period, ªloose gravelº and ªrunning driftº were
transported through the system.
Channel response was limited to bed oscillations
about a fairly constant mean bed height from 1961
to 1984 when active timber harvesting had progressed
further into the upland areas of the catchment. We
interpret the bed elevation ¯uctuations as pulses of
sediment in transport through the system. These bed
elevation oscillations increased in amplitude in the
1950s, a time in which the South Fork channel
width was increasing and harvesting was becoming
extensive in the headwaters. However, the bed did
degrade during this time of high-amplitude bed
S.C. Stover, D.R. Montgomery / Journal of Hydrology 243 (2001) 272±286282
0
0.5
1
1.5
2
2.5
1930 1940 1950 1960 1970 1980
Year
GH for 60 cms GH for 50 cms GH for 40 cms
GH for 30 cms GH for 10 cms GH for 8 cms
GH for 6 cms GH for 4 cms GH for 2 cms
Gag
eH
eigh
t(m
)
0
0.5
1
1.5
2
2.5
1930 1940 1950 1960 1970 1980
Year
GH for 60 cms GH for 50 cms GH for 40 cms
GH for 30 cms GH for 10 cms GH for 8 cms
GH for 6 cms GH for 4 cms GH for 2 cms
Gag
eH
eigh
t(m
)
Fig. 8. South Fork discharge/gage height relationship. Pro®le parallels the temporal changes in mean bed height, with an apparent stabilization
of the discharge/gage height trend with time. GH represents gage height corresponding to a given discharge (cms).
oscillations from 1949 to nearly 1961, likely due to
debris removal, in-channel perturbations and subse-
quent sediment mobilization. We interpret these
responses as recording a substantial transfer of sedi-
ment to the mainstem Skokomish River.
7.2. Mainstem
As with the South Fork, the timing of land manage-
ment activities and channel changes suggests that a
variety of in¯uences played a role in initiating the
trends observed in the mainstem bed elevation.
From 1932 to 1934, the mainstem exhibited substan-
tial and rapid degradation. During this time, Highway
101 was constructed less than 1 km downstream of the
gage, and heavy equipment was employed to extract
gravel from the mainstem. After gravel mining
ceased, the channel recovered some of the elevation
loss, but it was not until the ®rst noted logging in the
area in 1939 that the channel began to aggrade
rapidly. Aggradation continued until 1944, when
harvesting progressed upstream toward the head-
waters. From 1944 to 1964, the channel exhibited
enhanced bed height oscillations that we interpret as
pulses of sediment in transit through the system.
Because the appearance of these bed height oscilla-
tions corresponds to the time when channel width was
increasing for the South Fork, and upland harvesting
and road development became particularly extensive,
we infer that they re¯ect sediment input from channel
proximal disturbance and accelerated upstream sedi-
ment in¯ux. The hypothesis of an elevated sediment
supply to the mainstem is supported anecdotally by
the notes of USGS personnel from this time, such as
ªthis channel is a mess: there are bars and rif¯es
everywhere!º (USGS Discharge Sheet, September 4,
1957).
Comparison of the timing of net incision of the
S.C. Stover, D.R. Montgomery / Journal of Hydrology 243 (2001) 272±286 283
1.5
2
2.5
3
3.5
4
1930 1940 1950 1960 1970 1980 1990 2000
Year
Gag
eH
eig
ht
(m)
GH for 80 cms GH for 60 cms GH for 50 cms
GH for 40 cms GH for 30 cms GH for 20 cms
GH for 9 cms GH for 7 cms GH for 5 cms
Fig. 9. Mainstem discharge/gage height relationship. Pro®le parallels the temporal changes in mean bed height, with a marked increase in the
gage height for any given discharge regime over time. GH represents gage height corresponding to a given discharge (cms).
South Fork and aggradation of the mainstem (Fig. 10)
suggests the translation of sediment from the South
Fork downstream to the mainstem. For sediment
released by degradation of the South Fork and the
later sediment pulses in the 1950s to have caused
aggradation of the mainstem, a transport rate of
300±660 m/year would be required, which is similar
to other reported sediment wave propagation rates of
365±730 m/year (Goff and Ashmore, 1994), 600 m/
year (Grif®ths, 1979; 1993) and 1000 m/year
(Beschta, 1983).
In addition, mainstem aggradation may have been
S.C. Stover, D.R. Montgomery / Journal of Hydrology 243 (2001) 272±286284
Ele
vatio
nC
hang
e(m
)
Time Period (years)
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
South Fork
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
1932 - 1934 1934 - 1944 1944 - 1954 1954 - 1964 1964 - 1974 1974 - 1984 1984 - 1994
Mainstem
Ele
vatio
nC
hang
e(m
)
Time Period (years)
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
South ForkSouth ForkSouth Fork
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
1932 - 1934 1934 - 1944 1944 - 1954 1954 - 1964 1964 - 1974 1974 - 1984 1984 - 1994
MainstemMainstemMainstem
Fig. 10. 1932±1994 aggradation and degradation trends for the mainstem and South Fork reaches, derived from volumetric changes at each
reach using Eq. (2). With the exception of 1944±1954, the mainstem aggraded during the South Fork degradation periods.
in¯uenced by placement of riprap and natural berm
material along the right bank of the channel, upstream
of the gaging location beginning between 1971 and
1975 (USGS discharge sheets, various dates). These
constraints extend intermittently for nearly 1 km, and
served to enhance aggradation and further reduce the
mainstem's transport capacity by restricting the
ability of the channel to widen in response to
increased sediment loads.
7.3. Implications for monitoring programs
Many channel monitoring programs rely on
repeated surveys of channel cross-sections for detect-
ing changes in channel geometry. The distinct subann-
ual to multi-year variability apparent in mean bed
height trends for both the South Fork and mainstem
reaches shows that the interpretation of meaningful
trends in mean bed elevations will be strongly
impacted by measurement frequency; either frequent
measurements or a long record would be required to
detect meaningful trends in channel response. For
example, annual monitoring of the mainstem
Skokomish River from 1958 to 1964 or 1968 to
1976 would not have detected the dramatic ongoing
aggradation. Further attention to the design of
programs for detecting meaningful trends in channel
response from cross-sectional surveys is warranted,
especially given the dif®culty in sustaining long-
term monitoring programs.
8. Conclusions
Our analysis of gaging station records for the
Skokomish River indicates that long-term evaluation
of trends in both channel morphology and discharge
regimes are necessary when examining the effects of
land use on ¯ooding. The reduction in the requisite
discharge for overbank ¯ow, together with a reduction
in the peak discharge regime, shows that recent ¯ood-
ing along the mainstem Skokomish is due to aggrada-
tion within the channel. Without signi®cant
aggradation of the ¯oodplain, the higher stage
associated with smaller discharges allows for a higher
frequency of ¯oodplain inundation on the mainstem.
Although we have only limited data on land use
history, the timing of observed channel changes
implies a relationship between land uses and
downstream ¯ooding. We suggest that increased sedi-
ment loading in the South Fork, combined with a
reduced transport capacity from dam construction on
the North Fork, resulted in substantial aggradation on
the mainstem. The increased ¯ooding along the
Skokomish River due to loss of channel conveyance
illustrates the importance of considering sediment
transport processes and geomorphological change in
the assessment and management of ¯ood hazards in
forested mountain drainage basins.
Acknowledgements
We would like to thank the anonymous reviewers
for their useful comments that served to strengthen the
manuscript. This research was supported in part by the
Center for Streamside Studies at the University of
Washington. Luis Fuste and David Kresch of the
USGS Tacoma, Washington of®ce provided historic
discharge measurements and photography of the
Skokomish Basin. Bill Shelmerdine, Leslie Lingley,
and Kyle Noble from the USFS Olympia, Washington
branch provided aerial photographs and landslide
mapping data.
References
Beschta, R.L., 1983. Long-term changes in channel widths of the
Kowai River, Torlesse Range, New Zealand. Journal of Hydrol-
ogy (New Zealand) 22, 112±122.
Booth, D., 1991. Urbanization and the natural drainage system,
impacts, solutions, and prognoses. The Northwest Environmen-
tal Journal 7, 93±118.
Bosch, J.M., Hewlett, J.D., 1982. A review of catchment experi-
ments to determine the effect of vegetation changes on water
yield and evapotranspiration. Journal of Hydrology 55, 3±23.
Canning, D.J., Randlette, J.L., Haskins, W.A., 1988. Skokomish
River Comprehensive Flood Control Management Plan.
Washington Department of Ecology. Report 87-24, Olympia,
Washington.
Carson, R.J., 1970. Quaternary Geology of the South-Central Olym-
pic Peninsula, Washington. PhD dissertation. University of
Washington.
Goff, J.R., Ashmore, P., 1994. Gravel transport and morphological
change in braided Sunwapta River, Alberta, Canada. Earth
Surface Processes and Landforms 19, 195±212.
Grif®ths, G.A., 1979. Recent sedimentation history of the Waima-
kariri River, New Zealand. Journal of Hydrology (New Zealand)
18, 6±28.
Grif®ths, G.A., 1993. Sediment translation waves in braided gravel-
S.C. Stover, D.R. Montgomery / Journal of Hydrology 243 (2001) 272±286 285
bed rivers. Journal of Hydraulic Engineering Ð ASCE 119 (8),
924±937.
Harris, D.D., 1973. Hydrologic changes after clear-cut logging in a
small Oregon coastal watershed. Journal of Research of the US
Geological Survey 1, 487±491.
Hoey, T.B., 1994. Patterns of sediment storage in the Kowai River,
Torlesse Range, New Zealand. Journal of Hydrology (New
Zealand) 32, 1±15.
Hollis, G.E., 1975. The effect of urbanization on ¯oods of different
recurrence interval. Water Resources Research 11, 431±435.
James, L.D., 1965. Using a digital computer to estimate the effects
of urban development on ¯ood peaks. Water Resources
Research 1, 223±234.
Jay, D., Simestad, C.A., 1994. Downstream effects of water with-
drawal in a small, high-gradient basin: erosion and deposition on
the Skokomish River delta. Estuaries 17, 702±715.
Jones, J.A., Grant, G.E., 1996. Peak ¯ow responses to clear-cutting
and roads in small and large basins, western Cascades, Oregon.
Water Resources Research 32, 959±974.
KCM, 1993. Skokomish River Comprehensive Flood Hazard
Management Plan, vol. II, prepared for Mason County Depart-
ment of Community Development.
Lyons, J.K., Beschta, R.L., 1983. Land use, ¯oods, and channel
changes: Upper Middle Fork Willamette River, Oregon
(1936±1980). Water Resources Research 19, 463±471.
Macdonald, A., Keller, E.A., 1987. In: Beschta, R.L. (Ed.), Erosion
and Sedimentation in the Paci®c Rim. IAHS Publication No.
165, pp. 405±406.
Marsh, G.P., 1864. Man and Nature. Charles Scribner, New York
(472 pp.).
Moscrip, A.L., Montgomery, D.R., 1997. Urbanization, ¯ood
frequency, and salmon abundance in Puget lowland streams.
Journal of the American Water Resources Association 33,
1289±1297.
Rothacher, J., 1970. Increases in water yield following clear-cut
logging in the Paci®c Northwest. Water Resources Research
6, 653±658.
Rothacher, J., 1973. Does harvest in west slope Douglas-®r increase
peak ¯ow in small harvest streams? Res. Pap. PNW-163, US
For. Serv. For., Pac. Northwest For. and Range Exp. Stn.,
Corvallis, OR.
Shields, F.D., Smith, R.H., 1992. Effects of large woody debris
removal on physical characteristics of a sand-bed river. Aquatic
Conservation Ð Marine And Freshwater Ecosystems 2, 145±
163.
Smith, R.D., Sidle, R.C., Porter, P.E., Noel, J.R., 1993a. Effects of
experimental removal of woody debris on the channel morphol-
ogy of a forest, gravel-bed stream. Journal of Hydrology 152,
153±178.
Smith, R.D., Sidle, R.C., Porter, P.E., 1993b. Effects on bedload
transport of experimental removal of woody debris from a
forest, gravel-bed stream. Earth Surface Processes and Land-
forms 18, 455±468.
Surell, A., 1841. EÂ tude sur les Torrents des Hautes-Alpes. Dunrod,
Paris.
Tabor, R.W., 1975. Guide to the Geology of Olympic National Park.
University of Washington Press, Seattle (144 pp.).
Thomas, R.B., Megahan, W.F., 1998. Peak ¯ow responses to clear-
cutting and roads in small and large basins, western Cascades,
Oregon: a second opinion. Water Resources Research 34, 3393±
3403.
Williams, J.R., Pearson, H.E., Wilson, J.D., 1985. Stream¯ow
Statistics and Drainage-Basin Characteristics for the Puget
Sound Region, Washington, Vol. I: Western and Southern
Puget Sound. Open File Rep. 84-144-A, US Geological Survey,
Tacoma, Washington.
Ziemer, R.R., 1981. Storm ¯ow response to road building and
partial cutting in small streams in Northern California. Water
Resources Research 17, 907±917.
S.C. Stover, D.R. Montgomery / Journal of Hydrology 243 (2001) 272±286286